Kriging-based accelerated prediction of frequency spectrum and acoustic field around rotating sources

Abstract

The computational cost of predicting sound radiated from rotating sources is mainly affected by three parameters, i.e., the number of frequencies, source points and field points. The key of reducing the prediction cost of sound radiated from rotating sources is to reduce the times of integral calculation invoked in multi-frequency and combination of multi-source-field points. In view of the above engineering background, this paper employs a frequency-domain integral prediction method combined with a Kriging interpolation model to accelerate acoustic prediction. Three numerical cases are performed to validate the effectiveness of the above method in accelerating prediction of the frequency spectrum and acoustic field around rotating sources. The research also provides recommendations on the distribution characteristics of the initial sampling points and the method of adding sampling points.